AI is touted as the next big thing in almost every industry. From agriculture to transport, there’s barely any industry that is yet to explore at least an aspect of artificial intelligence and how it can enhance it in some way. In the context of eCommerce specifically, the applications of AI go from bots and virtual assistants to its most widespread application — assisted product discovery.
But what is product discovery and what role does AI play in simplifying it?
Simply put, product discovery is the experience of easily finding the products you want to buy. This process of discovery is spread across search or navigation — where the shopper expresses the intent to look for and subsequently purchases a particular product or through recommendations — where the shopper is shown the right product suggestions that match who they are and what would appeal to their sensibilities.
Traditionally, this mapping of products to phrases was a manual process. Online merchandisers are a lot like their retail counterparts, figuring out which products to promote, the display layout and sort order as well as planning and executing many seasonal and promotional campaigns.
The advancement of AI has made this process a lot more scientific.
Mixing up the order and display of products during seasonal promotions and encouraging diversity in the product display order for the same search term are other areas where AI-assisted merchandising come into play.
If your catalog is a maze, AI-assisted product discovery is like finding a map that makes it easier to manoeuver it.
Role of AI in Product Discovery
To the uninitiated, Daddy’s Car sounds a lot like any other song created by a pop band that’s had an overdose of the Beatles. The interesting bit is that the song, which has over 1.9 million views on YouTube, was composed using a system of machine learning algorithms called ‘Flow Machines’.
*let the jaws drop*
But after we’re done being amazed by what AI has made possible, it would be worthy to note that even in the case of Daddy’s Car, a human composer put together the melody created by the algorithm. Which brings us to a pertinent question — is AI, as it exists today, a creator or more of a collaborator?
There was a time when Goodreads helped me find the kind of books I’d enjoy reading. That was before Amazon got this smart. Almost every time I make a purchase on Amazon, it shows me a set of books that I can’t resist looking at. And though we’ve had a few misses, I’ve enjoyed most of those reads.
One of the pioneers of hyper personalization, Amazon has understood its shoppers over a period of time and now offers each one of them product discovery experiences that are unique to who they are.
This personalization isn’t limited to just recommendations. Personalization is essential across search and navigation journeys as well. You wouldn’t walk into a store and expect to be greeted by one single store associate to herd the entire set of shoppers through the store. Why else would a static experience make sense in the online space?
The future of online commerce is hyper personalized and focuses squarely on delighting the customer.
The Power of Human Insight
An algorithm trained to respond to a shopper’s likes and preferences, would have never been able to fathom the sudden interest in Uggs last year. The ugly, furry, uber-comfortable boots made an unexpected comeback towards the end of 2017 thanks to the likes of Sienna Miller and Rihanna. Not surprisingly, they made it to most fashion sites as well. Insight — the ability to unearth ideas, trends, and possibilities that lie hidden from plain sight — is something unique to the human mind. It is a frontier that AI is yet to climb since most of AI is still reliant on replicating what is has learnt from the huge data sets it ingests. It is merchandiser insight that identifies a relevant trend at the right time and capitalizes on it by showcasing it to an interested audience who is most likely to respond to the trend. Another classic online merchandising example where human insight plays a role over and above AI is for championing new products that recently added to the catalog. The unavailability of prior performance data for new products added to a catalog make it difficult to decide its place in the overall product order. In most situations, these products are added to the top or bottom of results making it almost impossible to be discovered. If all control lay with AI, it would continue to surface popular, trending items and never pull up the more recent additions. Adding a layer of human insight over this AI helps in attaching rules that aid in the discovery of these new products. At Unbxd, we handle this by assigning a ‘freshness score’ to every product in the catalog. New products get sufficient visibility during the initial days and the cumulative feedback from shoppers help decide their overall popularity and rank among the other products.